College grades are getting artificially inflated thanks to AI tools like ChatGPT, and it’s creating a headache for employers trying to figure out which graduates actually know their stuff. Students are increasingly using AI to boost their academic performance, but they might not be learning much in the process.
The trend has major implications for the job market. As entry-level positions become more competitive, companies are relying more heavily on GPAs to screen candidates. But if those grades don’t reflect genuine ability, the entire hiring process becomes less reliable.
A new study from the University of California, Berkeley revealed that courses heavy on writing and coding saw a significant spike in A grades after ChatGPT’s debut. Professors in these AI-exposed classes handed out about 30% more A’s while giving fewer A-minus and B-plus grades.
The research examined over half a million grades from 2018 to 2025 at a large Texas public university. Igor Chirikov, the study’s author and a senior researcher at Berkeley’s Center for Studies in Higher Education, found little difference between AI-exposed and non-AI-exposed courses through 2022. After that, A grades shot up specifically in classes where students could easily use AI tools.
This isn’t just about students getting better at their coursework. The results suggest students are using generative AI to improve their grades without actually learning more, according to Chirikov. Classes with more take-home assignments showed even higher rates of A grades, since students had more opportunities to use AI assistance.
The timing couldn’t be worse for the job market. While GPAs became less important for hiring in recent years, that trend is reversing as companies face huge numbers of entry-level applicants. In 2023, 37% of employers used grade-point averages in hiring decisions. That figure jumped to 42% by this year, with companies like Barclays and Morgan Stanley setting GPA minimums for internship roles.
The problem extends beyond just grades. AI is making traditional evaluation methods less reliable across the board:
- Cover letters can be AI-generated
- Resumes may include AI-assisted content
- Job application materials often use AI tools
- Academic work increasingly relies on AI assistance
Elite universities are starting to take notice. Harvard College released a February report noting that current grading policies don’t allow employers to properly compare student performance. Faculty there are voting on a proposal to cap the number of A’s awarded. Yale University was even more direct in an April report, stating that “grades exist to communicate what students have learned. At Yale, as at many peer institutions, they no longer do.”
The shift is creating a fundamental problem in education. Learning requires what Chirikov calls “productive struggle” that gets eroded when AI does the heavy lifting. Students might graduate able to produce polished reports and data analysis without developing critical thinking skills.
Companies are caught in a contradictory position. They dislike when students use AI in job applications but simultaneously want new hires who know how to use the technology effectively. “They’re talking from both sides of their mouth,” says Chelsea Schein, vice president of research strategy at Veris Insights, which analyzes hiring trends.
The standards for entry-level positions are getting tougher as a result. Among job postings that require GPAs, nearly a quarter now demand a 3.5 or higher, compared to just 9% in 2020. This creates additional pressure on students to maintain high grades, potentially encouraging even more AI use.
Some professors are adapting their teaching methods. Schein, who teaches at the University of Pennsylvania’s Wharton School, reduced the weight of homework in final grades after realizing AI could easily score 100% on her assignments. She now emphasizes in-class exams and quizzes where students can’t access AI tools.
The broader question is whether this technological shift fundamentally changes what education should measure. An A grade might now indicate that a student has access to better AI tools or knows how to use them more effectively, rather than demonstrating mastery of the underlying subject matter. This creates new challenges for both educators trying to assess learning and employers trying to identify qualified candidates.
